TEI Technical Council F2F Graz 2019-09-14/16

Venue

Centre for Information Modelling - Austrian Centre for Digital Humanities, University of Graz

Elisabethstraße 59/III, 8010 Graz

Saturday, Sep 14 09:00–12:30 Council meeting
14:00–17:00 Council meeting
Sunday, Sep 15 09:00–12:30 Council meeting
14:00–17:00 Council meeting
Monday, Sep 16 09:00–12:30 Council meeting (+invite people standing for Council)
14:00–16:00 Standoff meeting (Guests: Laurent Romary, Georg Vogeler, Lou Burnard, Piotr Banski)
17:00–18:00 Joint Board-Council Meeting
Thursday, Sep 19 19:00 Infrastructure Group Meeting

Present

Saturday, Sep 14 - Morning Session (9:00-12:30)

Saturday, Sep 14 - Afternoon Session (14:00–17:00)

Sunday, Sep 15 - Morning Session (09:00–12:30)

Sunday, Sep 15 - Afternoon Session (14:00-17:00)

Monday, Sep 16 - Morning Session (9:00–12:30)

Monday, Sep 16 - Afternoon Session (14:00–17:00)

Proposal 1:

<TEI>
   <teiHeader>
   <text>
   <standOff>

or Proposal 2:

<TEI>
   <teiHeader>
   <text>
   <TEI>
      <teiHeader>
      <standOff>
   <TEI>
      <teiHeader>
      <standOff>
   <TEI>
      <teiHeader>
      <standOff>

or

Proposal 3:

<TEI>
  <teiHeader>
  <text>
  <standOff>
     <standOffHeader>

Monday, Sep 16 - Joint Board-Council Meeting (17:00–18:00)

Stylesheets issues

Stylesheets pull requests

Group A: RV, EBB, MT Group B: MS, JC, MH, EM

Group C: HC, SS, PS, SB

Ticket Group Description Notes
#377 B Add instructions on how to build documentation PR merged MH asked Luis to fix two broken links in the documentation and to turn on directory browsing.
#369 C Improve error handling for faulty moduleRef uri F2F subgroup is just going to implement this one. It’s a minor change.
#327 B use ODT 'Author' style to specify author PS to implement. Green for Go.
#301 C Reordered stylesheets and addressed #281 PS has already addressed this difficult to discern PR on the PR itself. Main problem is we do not have sufficient CIDOC-CRM knowledge to evaluate some bits. Suggest that someone from Council ask Ontologies SIG for input.
#109 A Fix missing first text node, a [, in certain references. Fixed!

TEI Guidelines issues under discussion

Group A: JC, HC, EB, SB Group B: MH, MT, PS, MS

Group C: EM, RV, SS

Ticket Group Description Notes
#1813 A examples in ODD example page do not use ODD pure Nudge LB to finish this ticket. Check if he needs write access to this page, which isn’t in the Guidelines.
#1823 B teiCorpus: make guidelines and data model consistent Subgroup (from GitHub comment): Subgroup suggests changing the content model to exclude text from <teiCorpus> and prescribe one or more <TEI> | <teiCorpus> children Considered corrigible error, made green for PS to fix. (He plans to fix by changing content model of <teiCorpus> directly — we may wish to submit another ticket to factor <text> out of model.resourceLike at a later date.
#1824 C Guidelines: customized ODD examples based on teiCorpus? We think that Vanessa’s question about actual encoding may be handled within TEI, but at the same time, examples for customization (and other parts of the TEI) are very western and could benefit from a broader set of examples.
#1825 A Revision of TCW-20 re branching in TEIC GitHub Repo Prod EBB to do this.
#1826 B migrate to travis.com PS did the migration today. We just need to watch to see that everything goes OK. Close the ticket if the builds and notifications run well.
#1830 C <derivation> should allow <bibl> in its content model MS: I agree with LB that the information about the original text should be supplied in <sourceDesc> Council discussion: Is <sourceDesc> actually right for describing the origin of a translation? <textDesc> not used very much Group C - might be reasonable to have a place for bibliography, not <sourceDesc>, but external to <derivation>. Then again it can contain everything else, so why not <bibl> These elements can contain text but all the examples are empty, perhaps worth reconsidering how they (should) work? Related to this, these elements are contained within <profileDesc> which is not supposed to encode bibliographic aspects Finally <sourceDesc>  describes the source from which an electronic text was derived (so not what that text was derived from) Conclude - add example of this type to <sourceDesc>. Assign to MH
#1834 A Guidelines page links broken Only a problem in Jenkins. Changed to low priority, and added a TEI Build label. This is really about build processing. Also seems to be working Paderborn Jenkins.
#1838 C allow <g> in <zone>/<line> yes, and make zone and line xtext as noted by Lou.
#1842 A `setting` should contain `placeName` Subgroup notes that you can add this in locale. But larger group discussion ends up agreeing with Dario, and will add <placeName> and tweak examples to use it.
#1844 C Remove @type from att.interpLike and use att.typed is this part of the @type reconciliation? Is green so will implement.
#1847 A move free-standing minOccurs= and maxOccurs on elements to class? Go for green.
#1848 B Allow titlePage and other block-level elements in rdg We already allow div and <floatingText> so we’ve already gone down the slippery rabbit hole, so why not <titlePage> and <text>? Council says yes to <titlePage> and also to all its children. Go for green.
#1851 A msContents/msItem should be replaced in tei:object with something non-MS specific Subgroup recognize this has larger implications for more generalized description of objects. We *could* change this and Subgroup doesn’t know how yet. Full Council: SB notes that the content model of <msDesc> is problematic because you have to choose between paragraph description OR structured content, and you can’t do both. We don’t want to continue that problem in a non-MS specific model. JC:* Should we build a new model based on <msContents>? * Or move to an entirely new model with help from object ontologists?
Afternoon session - same groups
#1849 C Add more discussion of editorial practice to TC chapter We suggest double status: Go, then review.
#1843 B `dateline` should be a member of `model.pLike.front` Council subgroup agrees in principle with the equivalence between <byline> and <dateline>; where one is available, the other should be. Both are available at the end of <front>, but only <byline> is available at the beginning of <front>, which makes no sense. Green: add <dateline> to model.pLike.front, and add a good example showing both <byline> and <dateline> together, or get one from Dario.
#1837 B improve explanation of @defaultExceptions on schemaSpec MH considers @defaultException to be an ugly cludge to solve TEI’s own problem and would ideally see it automatically handled by schema-generating software. Council: We need to investigate how to process @defaultException in odd2rng Stylesheet, which currently throws bizarre ID errors over namespace issues. Figure it out, decide whether to get rid of @defaultException, and then write up the documentation called for in the ticket.
#1852 B Define semantics of witDetail without @target Subgroup tends to disagree with that and recommend to do this on a project level. Full Council: agrees that we should add prose that <witDetail> when a child of <app>, when a sibling of <rdg> or <lem>, without @target, should conventionally refer to the first preceding sibling witness (<rdg> or <lem>). Add prose that projects may change this with documentation.
#1854 C Deprecations Appendix is seriously ugly yes - either make list with sub-items for cell values (mobile compatible) or fix table, fix column heads, move “component type” to column 4, remove or clarify deprecation statement which looks like an element name, etc. (another suggestion: shorten to “withdrawn after”). Poke Hugh to define new look and tell MH, SB. (this is automatically generated, right? – yes)
#1856 A `<ab>` should be able to nest Subgroup agrees to disagree and punt this to full Council. Council: semantically nested block-level text perhaps has consequences for processing: forcing some to add constraints to prevent nestable ab’s when they don’t want them. Perhaps introduce a new element (<nb> for nesting block) to prevent this trouble? <nb> sounds too much like nota bene. Better name? <nestBlock> or just permit <ab> to self-nest anyway?
#1857 B `<egXML>` should be able to nest see comment in ticket
#1858 C <q> should be allowed in <byline> Subgroup thinks we should implement this the way that SB outlines in the comment and should mark in green. Although we may want to consider the implications of putting things like <floatingText> in <byline> Full group: To be fixed with a redesign of <q> (new ticket: #1918)
#1859 A <altIdent> should never be empty <altIdent> should only permit XML names, but it’s currently allowed in <category>, <taxonomy> and <joinGrp> (where perhaps <gloss> would be a better choice?). This is a corrigible error, but we should do a deprecation period anyway in case people are using this in the wrong way.
#1860 EM, MH, JC, MS Encoding RDF relationships in TEI (TEI+RDFa and alternatives) should be discussed by the workgroup on LOD, and ensure some Council members are involved with that.
#1861 C Feature Request: <listFigure> element Yes? Maybe? We aren’t opposed to the idea, but we aren’t sure if it needs its own element, or if <list> (in <sourceDesc>?) would suffice. Would this also be something that gets generated by processing (akin to <divGen>?) Full Council: action on EBB to work with Joey Takeda on some use cases for Council to consider. Look at <object> to see what it might need to refer to graphics the way <figure> does.
#1862 A Element content models with valList should not have any other content Sub-group thinks this constraint certainly belongs in p5odds (the schema P5 itself uses) and in tei_customization (the schema for writing TEI customization ODDs). We are unconvinced, however, that it belongs in P5 itself. If you want to define a markup language using ODD that allows this weird kind of data, power to you.
#1864 B Feature Request: Expand the content model of <catDesc> solution 3 implemented
#1868 C listWit should also allow msDesc and object Subgroup thinks that it would be best to keep things as they are so that <listWit> contains a list of <witness> elements, which can in turn describe witnesses using <msDesc>, <object>, etc. We don’t see an issue with the nesting that this requires.
#1869 A ODD spec elements should have their own source attribute Subgroup likes solution 2, and LB’s @lib attribute. (In large part because the obvious way to handle #1 doesn’t work here, because one of the attributes is global.) Full Council: We recommend calling it @specSource, but processing will have to support both @source and @specSource when pointing to an ODD source. We don’t know what elements get the special new attribute yet.
#1871 C restore explanation of TEI namespace JC: Note, best solution is probably to redirect the namespace URL to a different documentation URL, which is then https. (There saying it has been redirected.) F2F subgroup: agrees with JC HC: I don’t know how to do what JC suggests, but I have tried adding pages for the namespaces, and they seem to work. JC: No, they don’t work! They redirect to https which is completely wrong. What I’m saying is that the https://www.tei-c.org/ns/1.0 should redirect somewhere else precisely because we don’t want to confuse people with putting https in the namespace.
#1872 A Multiple sibling remarks in the same language cause translation issues Subgroup favours solution 2. Solution 1 may be more robust, but too much work to implement. This involves changing the stylesheets so that in a tagdoc the <remarks> are labeled “Notes” (plural). Full Council: Add schematron rule to say only one <remark> per language.
#1873 B Create Website Documentation Specifically Aimed at Developers Full Council: Maybe start working on this in the TEI wiki. Assign PS to make the space on the TEI wiki.
#1875 C sentence on tables merges style and rend attrs Good idea. Council approves if they agree and RV does it.
#1876 A att.pointing and att.scoping overlap Make att.pointing claim membership in att.scoping. Full Council: @match (with XPath) may be obsolete now that we have XPointer? @target takes you to known processible stuff. @match used to be use for uncertainty. Questions: Are we overcomplicating the class system in trying to reduce the number of permitted attributes. Or is it more daunting for the user to have to choose from among a long list of attributes CLOSED-WON’T-FIX. All simple solutions are worse than doing nothing.
#1878 C datatype of @name of <attRef> too loose yes.
#1879 A New element listNote as a wrapper for note elements Subgroup: Wait until standoff proposal figures out what’s going on with annotations
#1881 C Repair ambiguous language referring to figures, tables, and examples Should be set to go: should be cleaned up for clarity. Sarah wants to do this.
#1883 A simplePrint ODD has duplicate lines already fixed and closed
#1885 C Update teidata.temporal.iso (and teidata.duration.iso?) We should be conformant with ISO. So we should update our datatypes. Council: This needs more review to see how it affects our current practice of encoding ambiguous dates.
#1889 A why cant i define a PM in terms of model classes rather than individual elements? JC: Sounds like a great idea. Hard to implement. MT: Probably not that useful.If we had inheritance-based classes it might be easier. But it might not benefit users to have to study the model classes and their various dependencies.
#1902 C Add model.bibl to model.entryPart.top Not clear whether OP wants <bibl>, model.biblPart, or model.blblLike. SS to ask on ticket.
#1906 A repositories and manuscripts in multiple locations JC: Original Poster is mistaken in what they are doing. Council: JC will make a final response to close this ticket
#1907 B Datatype example doesn’t do what prose implies it does Sub-group (SB,HC,PS,SS) thinks that SB’s (2) is the better solution, and leaves it to assignee (EBB) to improve the wording.
#1909 A non-formulaic formula Closed
#1910 B Add location information to the new `<conversion>` element SB—Not sure I understand entirely, and I agree with @duncdrum’s request for examples. But I suspect an attribute that points to a <place> is better than a child <placeName>